Analysis of Bayesian Classifier Accuracy

Abstract

The naïve Bayes classifier is considered one of the most effective classification algorithms today, competing with more modern and sophisticated classifiers. Despite being based on unrealistic (naïve) assumption that all variables are independent, given the output class, the classifier provides proper results. However, depending on the scenario utilized (network structure, number of samples or training cases, number of variables), the network may not provide appropriate results. This study uses a process variable selection, using the chi-squared test to verify the existence of dependence between variables in the data model in order to identify the reasons which prevent a Bayesian network to provide good performance. A detailed analysis of the data is also proposed, unlike other existing work, as well as adjustments in case of limit values between two adjacent classes. Furthermore, variable weights are used in the calculation of a posteriori probabilities, calculated with mutual information function. Tests were applied in both a naïve Bayesian network and a hierarchical Bayesian network. After testing, a significant reduction in error rate has been observed. The naïve Bayesian network presented a drop in error rates from twenty five percent to five percent, considering the initial results of the classification process. In the hierarchical network, there was not only a drop in fifteen percent error rate, but also the final result came to zero.

A Framework for Optimal Worker Selection in Spatial Crowdsourcing Using Bayesian Network

“…Eventually, we obtain the high probable workers by excluding low probable workers through probabilistic predictions for each of the attribute. Furthermore, in BN, variables are not strongly correlated to each other, given the classification node [42]. In a complex environment as mentioned where results depend on various sources of information, may need to introduce or exclude any variable into the system for tuning system performance, it may still present accurate classification result in a large number of datasets by individually identifying and mapping each of the variables with the model classifier.…”

Integrated network models for predicting ecological thresholds: Microbial – carbon interactions in coastal marine systems

“…The NBN structure assumes independence between each taxonomical variable in the network (Flores et al, 2014). This NBN approach is considered a more simplistic, hence naïve, representation of environmental relationships than the cause and effect approach in defining the conditional structure of the model (Costa et al, 2013). Despite this simplified assumption NBN approaches have strong mathematical foundations and are effective in large, complex models with data limited conditions or for unstructured data (Li and Li, 2013;Xu and Ma, 2014).…”

Diagnosis of Cardiovascular Diseases with Bayesian Classifiers

“…While they enables efficient uncertainty reasoning with hundreds of variables, they also enables human experts to better understand the modelled domain. Felipe et al (2013) considered BN model with Naive Bayes algorithm is one of the most effective classification algorithms today, that competing with more modern and sophisticated classifiers. BN is a probabilistic model that consists of dependency structure and local probability.…”